Go-to-market roles are 49% of Mistral AI's open jobs, 34% of OpenAI's and 30% of Anthropic's, against 4% at xAI, where 54% of openings are data center jobs.
Share of each lab's open roles by department group: 1,978 postings on the public job boards of OpenAI, Anthropic, xAI and Mistral AI, collected Oct 10, 2026. Go-to-market covers sales, marketing, partnerships, customer-facing engineering and customer support. Colours mark the lab.
Source: OpenAI job board, Anthropic job board, xAI job board, Mistral AI job board, Google Careers, Epoch AI. Collected Oct 10, 2026 with the Ujeebu API. Chart: Ujeebu, CC BY 4.0.
Powered by UjeebuKey takeaways
- Go-to-market is the largest group of open roles: 594 of the 1,978 postings at the four labs (30.0%), ahead of engineering and product (423) and research (193).
- Mistral AI is the most sales-led: 104 of its 213 open roles (48.8%) are go-to-market, against 281 of 815 at OpenAI (34.5%) and 196 of 645 at Anthropic (30.4%). Epoch AI counted 28% at OpenAI and 31% at Anthropic in early March 2026, with its own grouping.
- xAI is hiring for data centers, not sales: 164 of its 305 open roles (53.8%) are data center construction, facilities and operations jobs, 163 of them in Memphis, Tennessee or Southaven, Mississippi, and only 13 (4.3%) are go-to-market.
- Research is a small slice everywhere: 64 open roles at OpenAI (7.9%), 67 at Anthropic (10.4%), 38 at xAI (12.5%) and 24 at Mistral AI (11.3%). Anthropic lists the most safety, policy and security roles, 106 (16.4%).
- OpenAI has the most open roles, 815, then Anthropic (645), xAI (305) and Mistral AI (213). Google DeepMind's listing on Google Careers showed 94 open roles on the same day.
The data
| Lab | Department group | Open roles | All open roles at the lab | Share of the lab's open roles |
|---|---|---|---|---|
| xAI | Compute and data centers | 164 | 305 | 53.8% |
| Mistral AI | Go-to-market | 104 | 213 | 48.8% |
| OpenAI | Go-to-market | 281 | 815 | 34.5% |
| Anthropic | Go-to-market | 196 | 645 | 30.4% |
| OpenAI | Engineering and product | 194 | 815 | 23.8% |
| Anthropic | Engineering and product | 153 | 645 | 23.7% |
| Mistral AI | Corporate functions | 41 | 213 | 19.2% |
| Anthropic | Safety, policy and security | 106 | 645 | 16.4% |
| xAI | Engineering and product | 45 | 305 | 14.8% |
| Anthropic | Corporate functions | 95 | 645 | 14.7% |
| Mistral AI | Engineering and product | 31 | 213 | 14.6% |
| OpenAI | Compute and data centers | 113 | 815 | 13.9% |
| xAI | Research | 38 | 305 | 12.5% |
| OpenAI | Corporate functions | 96 | 815 | 11.8% |
| Mistral AI | Research | 24 | 213 | 11.3% |
| Anthropic | Research | 67 | 645 | 10.4% |
| OpenAI | Safety, policy and security | 67 | 815 | 8.2% |
| OpenAI | Research | 64 | 815 | 7.9% |
| xAI | Safety, policy and security | 23 | 305 | 7.5% |
| xAI | Corporate functions | 22 | 305 | 7.2% |
| Anthropic | Compute and data centers | 28 | 645 | 4.3% |
| xAI | Go-to-market | 13 | 305 | 4.3% |
| Mistral AI | Compute and data centers | 8 | 213 | 3.8% |
| Mistral AI | Safety, policy and security | 5 | 213 | 2.3% |
Download the free sample CSV (CC BY 4.0) Full CSV, 24 rows
We make every effort to find high-quality data sources, but this data was extracted from the open web and we cannot guarantee its accuracy, completeness or reliability. Use it at your own risk.
How we got this data
We collected this data with the Ujeebu web scraping API, which fetches and renders public pages and returns the fields we need as structured data. The same API is available to you for your own collections.
Try it in the playground Free account, 5,000 credits, no card.
Methodology and known limits
What we collected. Every open role listed on the public job boards of four AI labs on Oct 10, 2026: OpenAI (its Ashby careers board), Anthropic and xAI (Greenhouse job board API) and Mistral AI (Ashby job posting API), collected with the Ujeebu web scraping API. For each posting we kept the posting ID, the department or team the lab files it under, and the location. 1,978 postings: OpenAI 815, Anthropic 645, xAI 305, Mistral AI 213. The Greenhouse totals were checked against each board's own job count.
How we counted. Each posting counts once, whatever the number of locations. The labs use their own department names (about 150 in all), so we mapped each one to six groups: go-to-market (sales, marketing, partnerships, solutions and forward-deployed engineering, customer support, government sales), engineering and product (software, product, design, data science, program management), research (research teams, science and human data or AI tutor roles), compute and data centers (compute, hardware, data center build and operations), safety, policy and security (safety, trust and safety, policy, security), and corporate functions (finance, legal, people, recruiting, IT, communications, workplace). The mapping uses the lab's department, never the job title. A bar is the group's share of that lab's open roles.
Google DeepMind and Meta. Google DeepMind posts its jobs on Google Careers, which shows a total (94 on Oct 10) but whose robots.txt allows only the first page of results, so we could not group DeepMind's roles by department. Meta has no public job board API, and its careers site's robots.txt says automated collection is prohibited without written permission, so Meta is not included.
Context. The March 2026 shares come from Epoch AI's analysis of lab job postings, not from our collection.
Known limits
- One day of postings, Oct 10, 2026. Postings show the roles a lab is trying to fill, not its headcount or how many people it will hire; one posting can hire several people, or none.
- Department groups follow each lab's own department names. Labs file similar jobs differently (Anthropic's Applied AI team is customer-facing and counts as go-to-market; OpenAI's Applied AI is its product engineering group), so a few points of share can move with the mapping.
- Google DeepMind (94 open roles on Google Careers) is not split by department, and Meta is not covered.
- Postings that a lab hides from its board, internal postings and recruiting agency listings are not counted.
- Epoch AI's March 2026 shares use its own grouping, so the comparison with our October shares is indicative only.
- No posting was dropped: all 1,978 had an ID and a department, with no duplicates.
We reviewed the sources' terms and robots.txt on Oct 10, 2026 before collecting. Spotted an error? Tell us.
Disclaimer: We make every effort to find high-quality data sources, but this data was extracted from the open web and we cannot guarantee its accuracy, completeness or reliability. Use it at your own risk.
Embed and cite
Free to reuse under CC BY 4.0 with a link back to this page.
Embed this chart
<iframe src="https://ujeebu.com/data/ai-lab-open-roles-by-department-2026/embed" width="100%" height="900" style="border:0" loading="lazy" title="The AI talent war: open roles at OpenAI, Anthropic, xAI and Mistral AI by department, October 2026"></iframe>
<p style="font-size:12px">Chart: <a href="https://ujeebu.com/data/ai-lab-open-roles-by-department-2026?utm_source=embed&utm_medium=referral&utm_campaign=data-ai-lab-open-roles-by-department-2026">"The AI talent war: open roles at OpenAI, Anthropic, xAI and Mistral AI by department, October 2026"</a> (CC BY 4.0). Powered by <a href="https://ujeebu.com/?utm_source=embed&utm_medium=referral&utm_campaign=data-ai-lab-open-roles-by-department-2026&utm_content=powered-by">Ujeebu</a></p>
Cite this
Ujeebu, "The AI talent war: open roles at OpenAI, Anthropic, xAI and Mistral AI by department, October 2026", https://ujeebu.com/data/ai-lab-open-roles-by-department-2026, updated Oct 10, 2026. CC BY 4.0.
PNG for LinkedIn (1080x1350) PNG for link previews (1200x630)
Is your brand in AI answers? Check it free.